BTW, the latest coverity run flagged a new issue in GeoFeatureGroupExtension::getScopedObjectsFrom() that obj is checked for NULL and later dereferenced - 'looks like there is a return statement missing in line 223.

Sorry, that was only half the answer - the attribute error is a side effect of the exception. The exception is unfortunately caught (in PathAreaOp.py) which causes the test to continue and run into that issue.

When calling the failing test explicitly with: FreeCADCmd.exe --run-test "PathTests.TestPathDressupDogbone.TestDressupDogbone.test02" it gibes this output: test02 (PathTests.TestPathDressupDogbone.TestDressupDogbone) Verify bones are correctly generated for a Profile. ... PathAreaOp.INFO:...

Looking further I found out that the curve behind the edge is a circle with a radius of 10, a normal of (0,0,1) but is located at the point (0,0,INF). I hope this helps to fix the code where the circle is created. That circle is Path.Area's workplane and gets it's Z value from the BoundBox of the i...

Looking further I found out that the curve behind the edge is a circle with a radius of 10, a normal of (0,0,1) but is located at the point (0,0,INF). I hope this helps to fix the code where the circle is created. That circle is Path.Area's workplane and gets it's Z value from the BoundBox of the i...

Looking further I found out that the curve behind the edge is a circle with a radius of 10, a normal of (0,0,1) but is located at the point (0,0,INF). I hope this helps to fix the code where the circle is created. That circle is Path.Area's workplane and gets it's Z value from the BoundBox of the i...

realthunder wrote:ping

It turns out that in the unsuccessful runs BRepLib_FindSurface in Area.cpp doesn't find the face (return from line 1026-1028). My OCC kung-fu is absolutely not strong enough to figure out what the actual problem is. Any help appreciated.

I can reproduce the dogbone error in Linux on OCC 7.2, not in other environments though. Interestingly the first run after booting the VM or building FV always succeeds and all subsequent runs with the same binaries fail.

I could reproduce the dogbone error in a Linux VM with OCC 7.2 environment last night. Currently rebuilding in Linux with OCE to see if that makes a difference. In the appveyor logs I also saw a cross link error which I didn't see in my linux tests. Dogbone itself has been stable for months, but thi...

Sorry for the hassle of needing to change the property, but there was no way for python to do this as simple as for c++. To keep backward cmpatibility best is to add some code to the restore finish callback for the relevant objects and change the property there if needed. That's good info, thanks v...